A new system called CW-Net was developed to predict when an autonomous car can make a mistake. This technology is able to convert the reasoning of the artificial intelligence that controls the vehicle into explanations understandable by humans.
The first tests carried out with the system showed that this information allows for better prediction of the vehicle's behavior. In this way, it is possible to anticipate problematic situations before they occur.
In addition, CW-Net is able to detect flaws that could otherwise go unnoticed by traditional monitoring systems. This represents a significant improvement in the safety of autonomous vehicles.
The development of this technology responds to the need to better understand the decisions made by artificial intelligence in autonomous cars, making the driving process more transparent and safe for all road users.
